5. Results for n and Q
Figs. 8 and 9 show the strong anti-correlation between n and Q. This has been observed and discussed by many authors (e.g. Smoot et al. 1992, Seljak & Bertschinger 1993, Lineweaver 1994). This anti-correlation has a simple explanation. In the plot, any increase of the slope lowers the y-intercept (at ) and any decrease of the slope raises the y-intercept. The anti-correlation is thus inherent with the use of the amplitude at as the normalization.
The increase of the size of the error bars on n and Q as h and are conditioned on and then marginalized can be seen by comparing Figs. 8 and 9. Our n and Q results are and K from Fig. 9 where both h and have been marginalized. Conditioning on , we get K (see also Fig. 6).
The four year COBE-DMR constraints on the amplitude and slope of the power spectrum at large angular scales are and K, and conditioning on , K (Bennett et al. 1996). These DMR results are in the context of , CDM models and they need to be corrected due to the mildly model-dependent extended tails of the Doppler peak which reach even into the low- region. After the correction, the DMR result becomes and K. Thus our results from a combination of recent CMB measurements in the context of critical density universes agrees well with the DMR-only result and reduces the error bars on both n and Q.
For the standard CDM model of Fig. 8 (, and ), we obtain CL) and K (95% CL). Using similar methods and a similar data set, several authors have reported similar results. de Bernardis et al. (1997) find CL). White et al. (1996) find ( CL). Hancock et al. (1997) find ( CL). The variations of these standard model n determinations are probably due to slightly different data sets, different treatments of the Saskatoon calibration and the details of the calculation.
© European Southern Observatory (ESO) 1998
Online publication: December 16, 1997